Dr L James Valverde, Jr

1
Towards a More Realistic Appraisal of Extreme
Weather Risk in the Northeast United States
Dr L James Valverde, Jr Director, Economics and
Risk Management Insurance Information
Institute 110 William Street New York, NY
10038 Tel [phone] Fax (212)
732-1916 jamesv_at_iii.org www.iii.org
5 May 2006
2
Presentation Outline

• Recent Assertions Concerning Hurricane Risk
  Exposure for New York City and Long Island
• Hurricane Risk in the Northeast United States
  Current Scientific Understanding
• A look at NOAAs hurricane return period
  estimates for NY City and Long Island
• Relevant background on the NOAA hurricane risk
  data
• What the NOAA hurricane risk data says and
  doesnt say
• Are hurricane return period estimates reliable
  predictors of future hurricane activity?
• Towards a More Robust and Realistic Appraisal of
  Extreme Weather Risk for New York City and Long
  Island
• Key components of a complete risk
  characterization
• Relevant history of extreme weather events in New
  York City
• Other potential hazards Northeasters

3
Recent Assertions Concerning Hurricane Risk
Exposure for New York City and Long Island

• In recent months, it has been asserted that
  concerns about hurricane risk exposure in New
  York City and Long Island are bogus
• In support of these claim, several NOAA
  statistics have been cited
• Recent statistics from the National Oceanic and
  Atmos-pheric Administration (NOAA) show that the
  odds of a category 4 or higher hurricane hitting
  New York City is once every five hundred years
  and on Montauk Long Island once every 130 years
  emphasis added
• On the basis of these values, some have concluded
  that the hurricane data hardly demonstrates a
  risk to New York City and Long Island

4
Hurricane Risk in the Northeast U.S. A Look at
NOAAs Return Period Estimates for NY City and
Long Island

• The 1-in-500 year and 1-in-130 year values
  for New York City and Long Island that have been
  cited in recent press releases are actually
  expected return periods they are not
  probabilistic projections of future hurricane
  activity for the region
• In simple terms, hurricane return period values
  are interpreted as the expected long-run
  frequency with which hurricanes of a specified
  intensity (e.g., Category 3, etc.) may be
  expected to occur within 75 NMi of New York City
  or Long Island
• Estimates of hurricane return periods are derived
  from historical data
• For example, a return period of 20 years for a
  Category 3 hurricane means that, on average,
  during the previous 100 years, a category 3
  hurricane passed within 75 miles of that location
  approximately 5 times

Source NOAA
5
Relevant Background on the NOAA Hurricane Risk
Data

• The NY City and Long Island hurricane risk data
  cited in recent press releases is produced by
  NOAAs National Hurricane Center Risk Analysis
  Program, HURISK
• HURISK estimates the expected return period for
  hurricanes of varying intensities
• Key elements of HURISKs estimation procedure
• A population of tropical cyclones falling within
  a 75mi circle of a given location (e.g., New York
  City) is obtained from NOAAs best-track file
• For that particular set of storms, the maximum
  wind within the 75mi circle is found
• A count is then conducted to determine how many
  systems had winds of 30-34 kts, 35-39 kts, etc.
• A probability distribution is then fit to this
  distribution of historical values (e.g., Weibel
  distribution)
• From the fitted distribution, the probability p
  of exceeding a specific wind speed is computed
• The return period value in years is simply the
  reciprocal, 1/p
• This value gives an indication of the number of
  systems that may occur in the future

Source NOAA
6
HURISK Results for Hurricanes Passing Within 75
NMi of New York City 1870 - 2004
Source Graph courtesy of Colin McAdie, NOAA
Tropical Prediction Center
7
NOAA Hurricane Risk Data in a More Realistic
Context
Source Based on data provided by the NOAA
Tropical Prediction Center
8
What the NOAA Hurricane Risk Data Says and
Doesnt Say

• The expected return period estimates for Category
  2 and Category 3 hurricanes passing within close
  proximity to New York City are commensurate with
  return period estimates for the Gulf Coast and
  the Southeast
• Estimates of expected return periods for Category
  4 and Category 5 hurricanes affecting New York
  City are compromised by limitations /
  deficiencies in the available historical record
  of hurricane activity for the region

9
Long-Run Return Period Estimates for Tropical
Cyclone Winds New York City, 1870 - 2004
Source Graph courtesy of Colin McAdie, NOAA
Tropical Prediction Center
10
Implications of the NOAA Long-Run Data

• Tropical cyclone winds consistent with Category 4
  and Category 5 hurricanes are not realized in
  NOAAs long-run (10,000 years) simulations for NY
  City
• This result is purely an artifact of the
  available historical record
• This result can easily be misinterpreted to mean
  that such events are impossible for the region
• Characterizations like this of long-run
  likelihoods are highly-dependent on geographic
  location
• Miami is much more likely to be affected by
  Category 4 5 hurricanes a fact that is well
  reflected in the historical record
• As a consequence, the long-run return period data
  for that region of the country is more reliable
  than that for the Northeast U.S.
• Uncertainty in return period estimates is
  greatest for severe hurricanes (Category 4 5)
  affecting NY City and Long Island, as these
  extreme weather events are seen to occur much
  less frequently in the historical record

11
Are Hurricane Return Period Estimates Reliable
Predictors of Future Hurricane Activity?

• Hurricane return period estimates are based on
  historical data (e.g., historical tropical storm
  activity from 1851-2004)
• The data may be incomplete, missing, or
  unreliable
• The data may not be representative, or perhaps
  not large enough to yield statistically
  significant estimates and inferences
• Detailed historical records for hurricane
  activity in the Atlantic Basin is 100 years
• Complicates efforts to arrive at reliable return
  period estimates of storm-of-the-century
  effects
• Estimates of hurricane return periods presume
  that the future will look like the past
• If all things were equal
• Mean return period estimates are not truly
  predictive, but rather, conditionally indicative
  of what the future might bring, given what has
  happened historically in terms of extreme
  hurricane phenomena
• Most useful in preparedness and response
  contexts
• Structural design

12
Towards a More Robust and Realistic Appraisal of
Extreme Weather Risk for New York City and Long
Island

• Focusing solely on the potential frequency of
  occurrence and severity of future hurricane
  activity for the region is misleading
• It is important, also, to characterize and
  evaluate two other components that together
  with frequency and severity of occurrence
  comprise extreme weather risk
• The intrinsic vulnerability of a specific region
  or location (e.g., New York City and Long Island)
  to hurricanes and other extreme weather events
• Given these vulnerabilities, the potential losses
  both economic and insured that are likely to
  result from an extreme weather event

13
Hurricane Risk in Context

• Hurricane Risk in Context

14
Number of Major (Category 3, 4, 5) Hurricanes
Striking the US by Decade
1930s mid-1960s Period of Intense Tropical
Cyclone Activity
Mid-1990s 2030s? New Period of Intense Tropical
Cyclone Activity
10
Tropical cyclone activity in the mid-1990s
entered the active phase of the multi-decadal
signal that could last into the 2030s
Already as many major storms in 2000-2005 as in
all of the 1990s
Figure for 2000s is extrapolated based on data
for 2000-2005 (6 major storms Charley, Ivan,
Jeanne (2004) Katrina, Rita, Wilma
(2005)). Source Tillinghast from National
Hurricane Center http//www.nhc.noaa.gov/pastint.
shtm.
15
Average Annual Tropical Cyclone Insured Losses
(Top 10 States, Millions)
Distribution of Annual Losses
Normalized losses adjusted for inflation,
housing density, wealth and wind insurance
coverage, based on historical data for 100-year
period 1900-1999. Source Tillinghast-Towers
Perrin
16
Inflation-Adjusted U.S. Insured Catastrophe
Losses By Cause of Loss, 1985-2004¹
Insured disaster losses totaled 221.3 billion
from 1984-2004 (in 2004 dollars). After 2005
season, tropical cyclones will account for 45
of the total.
1 Catastrophes are all events causing direct
insured losses to property of 25 million or more
in 2004 dollars. Catastrophe threshold changed
from 5 million to 25 million beginning in 1997.
Adjusted for inflation by the III. 2 Excludes
snow. 3 Includes hurricanes and tropical storms.
4 Includes other geologic events such as volcanic
eruptions and other earth movement. 5 Does not
include flood damage covered by the federally
administered National Flood Insurance Program. 6
Includes wildland fires.
Source Insurance Information Institute estimates
based on ISO data.
17
Insured Loss Claim Count for Major Storms of
2005
Hurricanes Katrina, Rita, Wilma Dennis produced
a record 3.2 million claims
Property and business interruption losses only.
Excludes offshore energy marine losses. Source
ISO/PCS as of February 8, 2006 Insurance
Information Institute.
18
Top 10 Most Costly Hurricanes in US History
(Insured Losses, 2005)
Seven of the 10 most expensive hurricanes in US
history occurred in the 14 months from Aug. 2004
Oct. 2005 Katrina, Rita, Wilma, Charley, Ivan,
Frances Jeanne
Sources ISO/PCS and Insurance Information
Institute
19
Top 11 Insured Property Losses Worldwide,
1970-2005 (2005)
Five of the 11 most expensive disasters is world
history affected the US within the past 4 years.
All figures are for total losses across all
locations, not just US. Katrina losses are a
preliminary III estimate. Sources ISO/PCS Swiss
Re, Natural Catastrophes and Man-Made Disasters
in 2003, Sigma, no.1, 2004
20
Extreme Weather in the Northeast United States A
Look at New York City and Long Island

• Hurricane Risk in Context

21
Number of Hurricanes Directly and Indirectly
Affecting the Northeast United States Since 1900
Source New Hampshire Office of Emergency
Management
22
Relevant History of Extreme Weather Events in New
York City

• Extreme weather in New York City and surrounding
  regions is not without significant historical
  precedent
• Hurricane of 1821. Produced sea-level rise of
  13ft in one hour in what is now Battery Park City
• West Indian Monster of 1893. Triggered a 30ft
  storm surge that bore a destructive path through
  South Brooklyn and Queens
• Long Island Express of 1938. Severe tidal surge
  hit Bayport, LI, causing 690 deaths and much
  damage to property and infrastructure
• Hurricane Agnes, 1972. Caused record inland
  flooding across the Northeast. Prior to Hurricane
  Andrew in 92, the United States most costly
  natural disaster
• Tropical Storm Floyd, 1999. New York state and
  upstate NJ received 10-15 inches of rain during a
  24-hour period. Caused massive flash flooding in
  New York City.

Source climate.org
23
Long Island Express of 1938
Great New England Hurricane of 1938 aka Long
Island Express caused severe damage through much
of the Northeast, including Long Island
Source WeatherUnderground.com
24
Damage Caused by the Long Island Express
Hurricane of 1938

• 700 deaths 708 injured
• 4,500 homes, cottages, and farms destroyed
  15,000 damaged
• 26,000 destroyed automobiles
• 20,000 miles of electrical power and telephone
  lines downed
• 1,700 livestock and up to 750,000 chickens killed
  
• 2,610,000 worth of fishing boats, equipment,
  docks, and shore plants damaged or destroyed
• Half the entire apple crop destroyed at a cost of
  2 million

Source SUNY Suffolk http//www2.sunysuffolk.edu/m
andias/38hurricane/damage_caused.html
25
Catastrophe Loss Management

• Hurricane Risk in Context

26
Hurricanes and Tropical Storms Are Not the Only
Potential Hazards The Case of Northeasters

• In addition to hurricanes and tropical storms,
  northeasters also hold the potential to cause
  significant damage to New York City and its
  surrounding regions
• The damage they cause can be more widespread than
  for hurricanes
• They can inflict greater damage than hurricanes
  since they often maintain their strength over
  several tidal cycles at a particular location
• Relevant historical events include
• Blizzard of 1888. 40 inches of snow fell in New
  York City 400 people died.
• Northeaster of December 1992. Prompted the
  closing of all transpor-tation infrastructure
  caused significant beach and dune erosion
• Storm of the Century, March 1993. Killed 270
  people and damaged more than 12,000 homes on Fire
  Island economic costs exceeded 3B

Source climate.org
27
Probability of Major Hurricane Landfall (CAT 3,
4, 5) in 2006
Average over past century. Source Dr. William
Gray, Colorado State University, December 6, 2005.
28
Value of Insured Commercial Coastal Exposure
(2004, Billions)
New York Insured Commercial Coastal Exposure
Source AIR
29
Value of Insured Residential Coastal Exposure
(2004, Billions)
New York Insured Residential Coastal Exposure
Source AIR
30
Total Value of Insured Coastal Exposure (2004,
Billions)
Source AIR Worldwide
31
Global Warming and Extreme Weather Potential
Implications for the Northeast United States

• Many credible experts believe that hurricane
  activity in the Atlantic Ocean may be on the rise
• There is growing concern that a frequency shift
  has occurred in the past decade for Atlantic
  tropical system
• Long-term effects of the multi-decadal
  oscillation
• Periodic effects of El Nino and La Nina
• The Atlantic and Caribbean Oceans are currently
  in a heightened period of hurricane activity
  this activity increase may continue for several
  more decades
• The U.S. Global Change Research Program reports
  that the minimum and maximum temperatures in NY
  City ave shown a long-term warming trend during
  the past century
• Global climate models project that the regions
  temperatures could increase by about 14F by
  2030 and by about 510F by 2100
• Most scientists believe that a warmer atmosphere
  increases the potential for extreme weather
  events
• The latest trends and events suggest that,
  looking forward, hurricanes may be more
  destructive due to increases in frequency,
  intensity, duration, and coastal property
  exposure

32
The Need for Dynamic Hurricane Risk Estimation

• Most attempts to characterize hurricane and
  extreme weather risk are static, in that they do
  not attempt to estimate or model hurricane risks
  over time
• Loss estimate models do not account for changes
  in building inventory
• Number, locations, types, and vulnerabilities of
  buildings in a region vary with time
• Not taking these factors into account compromises
  the accuracy of loss estimates
• It also complicates efforts to reliably appraise
  the potential effectiveness of possible
  mitigation strategies
• Hurricane risk stakeholders need to work towards
  flexible and robust measures of risk and
  vulnerability
• Need to be able to compute changes in expected
  annual hurricane losses over time
• Need, also, to understand the key factors that
  influence and drive those changes

Sources See, e.g., V. K. Jain et al., Modeling
Changes in Hurricane Risk Over Time, Natural
Hazards Review, May 2005.
33
Summary and Conclusions

• Major hurricanes many with considerable
  destructive power have historically affected
  New York City and LI, and will continue to pose a
  ongoing threat to the region
• The region is highly vulnerable to many of the
  potential consequences that are associated with
  extreme weather events
• Sea-level rise
• Flooding
• Storm surge
• Coastal erosion and potential loss of wetlands
• A realistic appraisal of the risks that extreme
  weather poses to the region must look at
• Frequency and severity of occurrences
• Intrinsic vulnerability of the region
• Loss Estrimation

34
INSURANCE INFORMATION INSTITUTE ON-LINE
If you would like a copy of this presentation,
please give me your business card with e-mail
address